Tencent Unveils Upgraded LLM for Text-to-Image Generation, Open-Sourcing Code for Industry Benefit

Tencent has released an upgraded version of its large language model for text-to-image generation, making it open-source for enterprises and individuals. The move is expected to benefit the industry as a whole and create an open-source ecosystem for next-generation vision generation.

Tencent Unveils Upgraded LLM for Text-to-Image Generation

In a significant move, Chinese tech giant Tencent has released an upgraded version of its large language model (LLM) for text-to-image generation, making it open-source for enterprises and individuals. The eight-month-old Hunyuan large language foundation model has undergone a major upgrade, enhancing its overall performance by 20%.

“The complete source code of its text-to-image LLM has been released on US open-source platforms Hugging Face and Github to benefit the industry as a whole and build an open-source ecosystem for next-generation vision generation.”

The upgraded LLM employs the DiT model architecture, also used by OpenAI’s text-to-video tool Sora. With its primary database in Chinese, the tool can effectively and accurately understand Chinese-language commands.

Open-Source Ecosystem

By making the source code available, Tencent aims to create an open-source ecosystem for next-generation vision generation. This move is expected to benefit the industry as a whole, allowing individuals and enterprises to access the program’s code, modify or share its design, fix broken links, or scale up its capabilities.

Industry Impact

Since launching Hunyuan last September, Tencent has integrated its LLM into various business units, including Tencent Cloud, Tencent Games, and super app WeChat. The AI-powered tool has also been provided to over 20 media outlets and advertising firms to facilitate their work.

Competition in the AI Space

The launch of the upgraded version comes a day after Microsoft-backed OpenAI unveiled its newest GPT model, GPT-4o, capable of natural human-computer interaction across text, image, video, and audio. Open-source technologies have played a crucial role in facilitating China’s ability to improve its LLMs and catch up with OpenAI’s innovative generative AI tools.

Alibaba’s Move

Alibaba Group Holding, owner of the South China Morning Post, has also taken an aggressive move to give third-party developers access to its models after launching its self-developed Tongyi Qianwen, or Qwen, LLM last year. Alibaba Cloud has provided access to 76 Qwen text generation models on ModelScope and Hugging Face.

Financial Performance

Both Tencent and Alibaba reported better-than-expected profits in the first quarter of 2024. Shenzhen-based Tencent reported a 62% jump to 41.9 billion yuan (US$5.8 billion) in the first quarter, fueled by strong advertising revenue, marking its first quarterly profit growth since last June. Alibaba reported a 10% increase in profit to 79.7 billion yuan in the financial year through to the end of March, marking its most profitable year since 2021.

AI-generated image

Tencent’s Hunyuan LLM